A method for extracting wind farm signal features and denoising optimized data

Abstract

The present invention belongs to the field of data processing technology, and in particular, relates to a method for extracting wind farm signal features and denoising and optimizing data. It includes: step 1. obtaining wind farm measurement signal parameters; step 2. obtaining EEMD parameters in the measurement signal; step 3. establishing a Hilbert-Huang transform model of the white noise clustering empirical method according to the measured signal parameters; step 4. analyzing the obtained signal to determine the filter window length; step 5. sampling the analytical signal for the number of points m; step 6. setting the number of iterations K; step 7. calculating the derivative value and average value of each sampling point of the analytical signal; step 8. calculating the weight of each point in the analytical signal filter window to denoise each signal point; step 9. iterating K times; step 10. obtaining accurate wind farm analytical signal characteristic values. The present invention can effectively and accurately extract signal features and denoise optimization of wind farms, is easy to implement, and has broad market application prospects.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and in particular, relates to a method for extracting wind farm signal features and denoising optimized data, and more particularly to a method for adaptive denoising of measurement data by extracting features based on a Hilbert-Huang transform (HHT) of an empirical white noise clustering method (EEMD) and then performing Gaussian filtering. Background Art

[0002] The current development of the power system is mainly focused on the development of pollution-free renewable energy. The main ways to use new energy to generate electricity are: wind power generation, solar power generation, tidal power generation, fuel cells and other technologies, as well as advanced energy storage technologies such as supercapacitors and battery packs that match these technologies. Renewable energy generation has obvious advantages: low location requirements, flexible installation locations, and decentralized distribution, which can better adapt to the characteristics of decentralized resources; compared with the traditional model, it not only reduces the cost of energy transportation, but also reduces the power loss of transmission lines; the access of new energy can improve the reliability of power supply, ensure the minimum power supply and basic living needs of residents in local areas, and become an effective supplement to the large power grid. However, while new energy has the advantages of being pollution-free and renewable, it also has many disadvantages, such as unstable energy output and low density. The power generation of wind turbines and photovoltaic power generation is more obviously affected by the environment.

[0003] When new energy sites are connected to the power grid, they will also bring different degrees of impact hazards to the power system. For example, the power generated by wind turbines depends to a large extent on wind speed, but wind speed is constantly changing. Therefore, wind turbines connected to the grid must always adjust their control according to changes in wind speed, with the ultimate goal of obtaining the maximum wind energy output; solar power generation is mainly affected by external temperature and light intensity. The external temperature changes relatively slowly, but the light intensity may change dramatically, especially in an environment where the weather is not very sunny, photovoltaic power generation will also change dramatically. Therefore, intermittent new energy power generation systems represented mainly by wind power generation and photovoltaic power generation generally have random and volatile power outputs, which will not only have an adverse effect on the stability of the power system, but also pose a threat to the safe operation of the original power grid.

[0004] The performance evaluation, operation optimization and system modeling of current power plants require historical data under steady-state conditions, and currently field experiments are mainly used to obtain sample data. With the improvement of data acquisition technology and the rapid development of data mining methods, in-depth analysis and research on massive measured data has become an important solution for the diagnosis and identification of grid-related characteristics of new energy stations. A large amount of data information is stored in the online monitoring system of power plants, but because the load of power plants is in a changing state, the implementation of historical data is mixed with a large amount of non-steady-state operation data. Compared with traditional experimental methods, processing real-time measurement data has little actual impact and a short working cycle. It is easier to obtain actual data.

[0005] Traditional data processing still has the problem of only using empirical mode decomposition. When the data is not pure white noise, the empirical mode will have modal overlap and cannot accurately describe the characteristics of wind power plant signal data. Summary of the invention

[0006] In view of the shortcomings of the above-mentioned prior art, the present invention provides a method for extracting wind farm signal features and optimizing denoising data. The purpose is to achieve more effective and accurate wind farm signal feature extraction and denoising optimization, and to provide a technical basis and practical method for wind farm data processing.

[0007] The technical solution adopted by the present invention to achieve the above-mentioned purpose is:

[0008] A method for extracting wind farm signal features and denoising optimized data comprises the following steps:

[0009] Step 1. Obtain wind farm measurement signal parameters;

[0010] Step 2. Obtain EEMD parameters in the measurement signal;

[0011] Step 3. Establishing a Hilbert-Huang transform model of the white noise clustering empirical method according to the measured signal parameters;

[0012] Step 4. Determine the filter window length according to the signal obtained by analyzing the Hilbert-Huang transform model;

[0013] Step 5. Sample the analytical signal at the number of points m;

[0014] Step 6. Set the number of iterations K;

[0015] Step 7. Calculate the derivative value and average value of each sampling point of the analytical signal;

[0016] Step 8. Calculate the weight of each point in the analytical signal filter window to denoise each signal point;

[0017] Step 9. Iterate K times;

[0018] Step 10. Obtain accurate wind farm analytical signal characteristic values.

[0019] Furthermore, the step 1 of obtaining the wind farm measurement signal parameters refers to obtaining the wind farm voltage, short-circuit capacity, frequency, and inertia detection parameters.

[0020] Furthermore, the step 2 of obtaining EEMD related parameters in the measurement signal includes:

[0021] (1) Obtaining the values ​​of two extreme points of the signal curve refers to determining at least two extreme points in the measured signal curve, namely, a maximum value and a minimum value;

[0022] (2) Obtaining the characteristic time length value of the signal curve, which refers to the time interval between adjacent extreme value points in the measured signal curve;

[0023] (3) Obtaining the inflection point value of the signal curve means that when the measured signal curve has no extreme value point but only an inflection point, first obtain the inflection point data and then obtain the extreme value point by differentiation.

[0024] Furthermore, in step 3, a Hilbert-Huang transform model of the white noise clustering empirical method is established according to the signal parameters obtained in step 2;

[0025] Step (1) determining the condition for selecting the characteristic function MF;

[0026] Step (2) adding white noise to the measured data;

[0027] Step (3) draw the upper and lower envelopes of the signal and calculate the mean difference h1;

[0028] Step (4) repeatedly screens the mean difference h1 using the characteristic function value;

[0029] Step (5) calculating the difference signal r1;

[0030] Step (6) repeatedly screening the difference signal r1 until no MF can be screened out;

[0031] Step (7) obtains the analytical signal.

[0032] Furthermore, the condition for determining the selection of the characteristic function MF in step (1) is that the characteristic function MF must satisfy the condition that the number of times the number of zero crossings in the acquired signal is equal to the number of extreme value points or differs by 1;

[0033] Step (2) adding white noise to the measured data means that uniform white noise is evenly distributed over the entire signal time, and different signal scales are automatically mapped to scales related to the white noise;

[0034] Step (3) draws the upper and lower envelopes of the signal and calculates the mean difference h1: it means using a cubic spline curve to connect the maximum and minimum values ​​of the processed signal to form an envelope, and calculating the average value m1 of the two and subtracting it from the measured data to obtain the mean difference h1:

[0035] h1=f(t)-m1

[0036] Where m1 is the average value of the maximum and minimum values, and f(t) is the original measurement signal;

[0037] Step (4) repeatedly screens the mean difference h1 through the characteristic function value: refers to taking the difference h1 as a new signal and performing the calculation in step (3) again, and iterating multiple times until h k If MF is satisfied, the obtained signal contains high-frequency components;

[0038] The calculated difference signal r1 in step (5) refers to the difference between the original measured signal f(t) in step (3) and the high-frequency component h k The difference is expressed as:

[0039] r1=f(x)-h1

[0040] Step (6) repeatedly screening the difference signal r1 until MF cannot be screened out: refers to taking the difference signal r1 in step (5) as the signal again and repeating the above step (5) until the obtained signal is a monotonic function, and its formula is:

[0041] r n =r n-1 -h n

[0042] Among them, r n is the calculated difference signal after the nth iteration, r n-1 is the calculated difference signal after the n-1th iteration, h n is the high frequency component of the nth iteration, n is the number of iterations;

[0043] Step (7) obtains the analytical signal: after the above processing, the original measured signal is mathematically expressed as the sum of n MF components and a residual term, and the formula is:

[0044]

[0045] Where f(t) is the original measurement signal, j is the jth iteration, n is the number of iterations, and h j (t) is the mean difference of the jth iteration, r n (t) is the calculated difference signal after the nth iteration.

[0046] Furthermore, step 4 determines the filter window length according to the analytical signal obtained in step (7): it means that the window length is pre-determined for the convenience of subsequent sampling, and the filter window length is 2M+1, where M is a non-zero integer.

[0047] Furthermore, in step 5, the number of sampling points of the analytical signal is counted as m; it means that the obtained analytical signal is sampled according to the size of the feature window, and the number of sampling points is recorded as m.

[0048] Furthermore, the calculation of the derivative value and average value of each sampling point of the analytical signal in step 7 refers to taking the derivative of the sampling points of the analytical signal and obtaining the average value, and the formula is:

[0049]

[0050] Among them, f′ k (m) is the average derivative of the analytical signal at the mth point of the kth iteration, f k (m+1) is the value of the analytical signal at the m+1th point in the kth iteration, f k (m-1) is the value of the analytical signal at the m-1th point in the kth iteration.

[0051] Furthermore, step 8 calculates the weight of each point in the analytical signal filtering window to denoise each signal point; it means setting the weight ω k (m+j,σ), where: is ω k is the weight of the kth iteration, m is the sampling point, j is the point of the window with unit length, σ is the standard deviation of the Gaussian function, which is a constant parameter. Its formula is a, and it is obtained through multiple operations when ω k (m,σ) is large, that is, σ>f′ k (m) After several iterations, the denoising effect is obvious but the mutation points will be smoothed out; when ω k (m,σ) is small, that is, σ<f′ k (m) After several iterations, denoising cannot be achieved. After obtaining the weights, the signal is denoised using formula b:

[0052] a:

[0053] b:

[0054] Where: k (m,σ) is the weight at sampling point m in the kth iteration, f k+1 (m) is the output value after the k+1th iterative filtering, exp is the exponential function with e as the base, and M is the window length.

[0055] A computer storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of a method for extracting wind farm signal features and denoising and optimizing data are implemented.

[0056] The present invention has the following beneficial effects and advantages:

[0057] The present invention provides a method for extracting wind farm signal features and denoising optimized data, taking into account the non-stable state quantity when acquiring wind farm data, and analyzing a large number of non-fundamental frequency transient signals contained in the signal by considering the wind farm implementation measurement signal data. The key information and data external characteristics in the measured signal are quickly extracted, and the data is de-noised. The signal characteristics can be extracted more effectively and reliably, and a stable and accurate signal can be obtained, providing a technical basis and a practical method for extracting relevant data from wind farm stations.

[0058] The present invention extracts features from measured data based on Hilbert-Huang transform (HHT) of white noise empirical clustering method (EEMD), and removes noise through an adaptive method of Gaussian filtering to obtain accurate data, thereby significantly improving the reliability of wind farm data optimization processing.

[0059] The method of the present invention is easy to implement. On the basis of the original empirical mode decomposition, the white noise clustering empirical method is added, and the adaptive denoising combined with Gaussian filtering is easy to implement in theory. At the same time, each prediction function has a ready-made algorithm, and the control strategy is also easy to implement. The method of the present invention is also very convenient for commercial development. With the increase in wind farms, there will inevitably be a large demand for this method to process wind farm measurement signal data, and the present invention has a good commercial development prospect. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0061] Figure 1 It is an overall flow chart of the method for extracting wind farm signal features and denoising and optimizing data of the present invention;

[0062] Figure 2 It is a schematic diagram of five mode simulations of characteristic decomposition of wind farm measurement signals according to the present invention;

[0063] Figure 3 is the error of the characteristic decomposition of the wind farm measurement signal by the present invention;

[0064] Figure 4 It is a simulation comparison diagram before and after Gaussian filtering denoising of the present invention. DETAILED DESCRIPTION

[0065] In order to more clearly understand the above-mentioned objectives, features and advantages of the present invention, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0066] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0067] Refer to the following Figure 1-Figure 4 The technical solutions of some embodiments of the present invention are described.

[0068] Example 1

[0069] The present invention provides an embodiment, which is a method for extracting wind farm signal features and denoising and optimizing data. Figure 1 The figure shows the overall flow chart of the method for extracting wind farm signal features and optimizing data by denoising. After obtaining the wind farm measurement signal, the Hilbert-Huang transform model of the white noise clustering empirical method is established to obtain the analytical signal. The analytical signal is then sampled and the signal stability value is obtained by the Gaussian filter adaptive method. Figure 1 It can be seen from the process in that the method of the present invention is divided into two major steps, one is to extract the feature of measurement signal parameters, and the other is to remove the noise of the measurement signal. The first is to extract the feature of the measurement data based on the Hilbert-Huang transform (HHT) of the white noise clustering empirical method (EEMD), and the second is to use the adaptive denoising method of Gaussian filtering. This is the essential difference from other methods.

[0070] The EEMD refers to the empirical white noise clustering method as the prior art.

[0071] The present invention discloses a method for extracting wind farm signal features and denoising optimized data. The signal feature extraction refers to the present invention using an EEMD method to overcome the problem that local defects of measurement data lead to the inability to accurately analyze the signal, and then removing noise by Gaussian filtering to obtain accurate signal data.

[0072] The present invention specifically comprises the following steps:

[0073] Step 1. Obtain wind farm measurement signal parameters;

[0074] The obtaining of wind farm measurement signal parameters refers to obtaining wind farm voltage, short-circuit capacity, frequency, and inertia detection parameters.

[0075] Step 2. Obtain EEMD parameters in the measurement signal;

[0076] It means that the measured signal parameters must have the following three points:

[0077] (1) Obtaining the values ​​of two extreme points of the signal curve refers to determining at least two extreme points in the measured signal curve, namely, a maximum value and a minimum value;

[0078] (2) Obtaining the characteristic time length value of the signal curve refers to the time interval between adjacent extreme value points in the measured signal curve;

[0079] (3) Obtaining the inflection point value of the signal curve means that when the measured signal curve has no extreme value point but only the inflection point, the inflection point data is first obtained and then the extreme value point is obtained by differentiation;

[0080] Step 3. Establish a Hilbert-Huang transform model of the white noise clustering empirical method according to the signal parameters obtained in step 2;

[0081] The following steps are involved:

[0082] Step (1) determining the condition for selecting the characteristic function MF; the characteristic function MF must satisfy that the number of times the zero point is crossed in the acquired signal is equal to the number of extreme value points or differs by 1;

[0083] The MF condition refers to the intrinsic mode function component, which as a characteristic function has the characteristic that the number of times it crosses zero points is equal to or differs by 1 from the number of extreme points.

[0084] Step (2) adds white noise to the measured data; this means that uniform white noise is evenly distributed over the entire signal time. Different signal scales will be automatically mapped to scales related to white noise.

[0085] Step (3) draw the upper and lower envelopes of the signal and calculate the mean difference h1;

[0086] It means using a cubic spline curve to connect the maximum and minimum values ​​of the processed signal to form an envelope, and calculating the average value m1 of the two and subtracting it from the measured data to get the mean difference h1:

[0087] h1=f(t)-m1

[0088] Where m1 is the average value of the maximum and minimum values, and f(t) is the original measurement signal.

[0089] Step (4) repeatedly screens the mean difference h1 using the characteristic function value;

[0090] It means taking the difference h1 as the new signal and performing the above step (3) again, iterating multiple times until h k If MF is satisfied, the obtained signal contains high-frequency components;

[0091] Step (5) calculates the difference signal r1; this is the difference between the original measured signal f(t) in step (3) and the high-frequency component hk The difference is expressed as:

[0092] r1=f(x)-h1

[0093] Step (6) repeatedly screens the difference signal r1 until MF cannot be screened out; it means that the difference signal r1 in step (5) is used as the signal again to repeat the above step (5) until the obtained signal is a monotonic function, and its formula is:

[0094] r n =r n-1 -h n

[0095] Among them, r n is the calculated difference signal after the nth iteration, r n-1 is the calculated difference signal after the n-1th iteration, h n is the high frequency component of the nth iteration, and n is the number of iterations.

[0096] Step (7) obtains the analytical signal;

[0097] It means that after the above processing, the original measured signal is mathematically expressed as the sum of n MF components and a residual term. The formula is:

[0098]

[0099] Where f(t) is the original measurement signal, j is the jth iteration, n is the number of iterations, and h j (t) is the mean difference of the jth iteration, r n (t) is the calculated difference signal after the nth iteration.

[0100] As shown Figure 2 As shown in FIG. 1 , there are five modal simulation schematic diagrams of the wind farm measurement signal feature decomposition of the present invention, such as Figure 3 FIG. 1 shows the error of the characteristic decomposition of the wind farm measurement signal according to the present invention.

[0101] Step 4. Determine the filter window length according to the analytical signal obtained in step (7);

[0102] Determining the filter window length refers to predetermining the window length for subsequent convenient sampling, and the filter window length is 2M+1, where M is a non-zero integer;

[0103] Step 5. Count the number of sampling points of the analytical signal as m;

[0104] The number of sampling points of the analytical signal refers to sampling the obtained analytical signal according to the size of the feature window, and the number of sampling is denoted by m.

[0105] Step 6. Set the number of iterations K;

[0106] The setting of the number of iterations K refers to initializing the data to K=0 and setting K iterations.

[0107] Step 7. Calculate the derivative value and average value of each sampling point of the analytical signal;

[0108] The calculation of the derivative value and average value of each sampling point of the analytical signal refers to taking the derivative of the sampling point of the analytical signal and obtaining the average value, and the formula is:

[0109]

[0110] Among them, f′ k (m) is the average derivative of the analytical signal at the mth point of the kth iteration, f k (m+1) is the value of the analytical signal at the m+1th point in the kth iteration, f k (m-1) is the value of the analytical signal at the m-1th point in the kth iteration.

[0111] Step 8. Calculate the weight of each point in the analytical signal filtering window to denoise each signal point;

[0112] The calculation of the weights of each point in the filter window of the analytical signal to remove noise from each signal point means: let the weight ω k (m+j,σ), where: is ω k is the weight of the kth iteration, m is the sampling point, j is the point of the window with unit length, and σ is the standard deviation of the Gaussian function, which is a constant parameter.

[0113] The formula is a, and after multiple calculations, we can get k (m,σ) is large, that is, σ>f′ k (m) After several iterations, the denoising effect is obvious but the mutation points will be smoothed out; when ω k (m,σ) is small, that is, σ<f′ k After several iterations (m), denoising cannot be achieved. After obtaining the weights, the signal is denoised using formula b:

[0114] a:

[0115] b:

[0116] Where: k (m,σ) is the weight at sampling point m in the kth iteration, f k+1 (m) is the output value after the k+1th iterative filtering, exp is the exponential function with e as the base, and M is the window length.

[0117] like Figure 4 As shown, it is a simulation comparison diagram before and after Gaussian filtering denoising of the present invention.

[0118] Step 9. Iterate K times;

[0119] The number of iterations K means that the denoising process needs to be iterated multiple times. When the number of iterations meets the output result set in step 6, the number of iterations K = K;

[0120] Step 10. Obtain accurate wind farm analytical signal characteristic values.

[0121] Example 2

[0122] Based on the same inventive concept, an embodiment of the present invention further provides a computer storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of a method for extracting wind farm signal features and denoising and optimizing data described in Example 1 are implemented.

[0123] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0124] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0125] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0126] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0127] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method for extracting wind farm signal features and denoising optimized data, characterized in that: The following steps are involved: Step 1. Obtain wind farm measurement signal parameters; Step 2. Obtaining EEMD parameters in the measured signal, including: (1) obtaining the values ​​of two extreme points of the signal curve, which refers to determining at least two extreme points in the measured signal curve - a maximum value and a minimum value; (2) obtaining the value of the characteristic time length of the signal curve, which refers to the time interval between adjacent extreme points in the measured signal curve; (3) obtaining the value of the inflection point of the signal curve, which refers to first obtaining the inflection point data and then obtaining the extreme point by differentiation when the measured signal curve has no extreme point but only an inflection point; Step 3. Establishing the Hilbert-Huang transform model of the white noise clustering empirical method based on the measured signal parameters, including: Step (1) determining the conditions for selecting the characteristic function MF; Step ... 2) adding white noise to the measured data; step (3) drawing the upper and lower envelopes of the signal and calculating the mean difference h1; step (4) repeatedly screening the mean difference h1 through the characteristic function value; step (5) calculating the difference signal r1; step (6) repeatedly screening the difference signal r1 until the MF cannot be screened out; step (7) obtaining the analytical signal; step 4. determining the filter window length according to the signal obtained by the Hilbert-Huang transform model; step 5. sampling the analytical signal at the number m of sampling points; step 6. setting the number of iterations K; step 7. calculating the derivative value and the average value of each sampling point of the analytical signal, which means taking the derivative of the sampling points of the analytical signal and obtaining the average value, and the formula is: Among them, f′ k (m) is the average derivative of the analytical signal at the mth point of the kth iteration, f k (m+1) is the value of the analytical signal at the m+1th point in the kth iteration, f k (m-1) is the value of the analytical signal at the m-1th point of the kth iteration; Step 8. Calculate the weight of each point in the analytical signal filter window to denoise each signal point, which means setting the weight ω k (m+j,σ), where: is ω k is the weight of the kth iteration, m is the sampling point, j is the point of the window with unit length, σ is the standard deviation of the Gaussian function, which is a constant parameter, and its formula is a. Through multiple calculations, we can get k (m,σ) is large, that is, σ>f′ k (m) After several iterations, the denoising effect is obvious but the mutation points will be smoothed out; when ω k (m,σ) is small, that is, σ<f′ k (m) After several iterations, denoising cannot be achieved. After obtaining the weights, the signal is denoised using formula b: a: b: Where: k (m,σ) is the weight at sampling point m in the kth iteration, f k+1 (m) is the output value after the k+1th iterative filtering, exp is an exponential function with e as the base, and M is the window length; Step 9. Iterate K times; Step 10. Obtain the accurate wind farm analytical signal characteristic value.

2. The method for extracting wind farm signal features and denoising and optimizing data according to claim 1 is characterized by: Step 1 of obtaining wind farm measurement signal parameters refers to obtaining wind farm voltage, short-circuit capacity, frequency, and inertia detection parameters.

3. The method for extracting wind farm signal features and denoising and optimizing data according to claim 1 is characterized by: The condition for determining and selecting the characteristic function MF in step (1) is: the characteristic function MF must satisfy that the number of times the zero point is crossed in the acquired signal is equal to the number of extreme value points or differs by 1; Step (2) adding white noise to the measured data means that uniform white noise is evenly distributed over the entire signal time, and different signal scales are automatically mapped to scales related to the white noise; Step (3) draws the upper and lower envelopes of the signal and calculates the mean difference h1: it means using a cubic spline curve to connect the maximum and minimum values ​​of the processed signal to form an envelope, and calculating the average value m1 of the two and subtracting it from the measured data to obtain the mean difference h1: h1=f(t)-m1 Where m1 is the average value of the maximum and minimum values, and f(t) is the original measurement signal; Step (4) repeatedly screens the mean difference h1 through the characteristic function value: refers to taking the difference h1 as a new signal and performing the calculation in step (3) again, and iterating multiple times until h k If MF is satisfied, the obtained signal contains high-frequency components; The calculated difference signal r1 in step (5) refers to the difference between the original measured signal f(t) in step (3) and the high-frequency component h k The difference is expressed as: r1=f(x)-h1 Step (6) repeatedly screening the difference signal r1 until MF cannot be screened out: refers to taking the difference signal r1 in step (5) as the signal again and repeating the above step (5) until the obtained signal is a monotonic function, and its formula is: r n ＝r n-1 -h n Among them, r n is the calculated difference signal after the nth iteration, r n-1 is the calculated difference signal after the n-1th iteration, h n is the high frequency component of the nth iteration, n is the number of iterations; Step (7) obtains the analytical signal: after the above processing, the original measured signal is mathematically expressed as the sum of n MF components and a residual term, and the formula is: Where f(t) is the original measurement signal, j is the jth iteration, n is the number of iterations, and h j (t) is the mean difference of the jth iteration, r n (t) is the calculated difference signal after the nth iteration.

4. The method for extracting wind farm signal features and denoising and optimizing data according to claim 1 is characterized by: Step 4 determines the filter window length according to the signal obtained by analyzing the Hilbert-Huang transform model: refers to predetermining the window length for the convenience of subsequent sampling, and the filter window length is 2M+1, where M is a non-zero integer.

5. The method for extracting wind farm signal features and denoising and optimizing data according to claim 1 is characterized by: The number of sampling points m for the analytical signal in step 5 means that the obtained analytical signal is sampled according to the size of the feature window, and the number of sampling points is recorded as m.

6. A computer storage medium, characterized in that: The computer storage medium stores a computer program, and when the computer program is executed by a processor, the steps of a method for extracting wind farm signal features and denoising and optimizing data as described in any one of claims 1 to 5 are implemented.

Images